Insured Profiles Segmentation using Unsupervised Machine Learning

Abstract

The National Social Insurance Fund (CNAS) in Algeria manages a crucial pillar of social security against risks such as illness or disability, covering salaried workers through various benefits. This study applies the K-means algorithm, an unsupervised machine learning method, to the Big Data from the General Directorate of CNAS, which is processed using the scikit-learn package in Python. to segment the insured into eight homogeneous profiles based on age, the frequency of prescribed medication care, and the amounts of reimbursements. The segments identified by the execution of the algorithm reveal a behavioral diversity in healthcare consumption, providing the CNAS with a basis for targeted actions in prevention, awareness, and optimization of reimbursements.